Healthy values in IOTHREADS?

[Former Member]

We're encountering severe problems with our MaxDB instances and want to define thresholds in order to convince our hoster to move data volumes from NFS to SAN devices. In peak times, WriteTime goes up to 2000 ms which is not acceptable for us. Which values for ReadTime / WriteTime are healthy?

[Former Member]

Hi Daniel,

this is a difficilt to answer questions because it depends a lot from the activities on the system.

Them IOm write time on the Log volume should be not more than 5 msec better would be < 5 msec
The read time on data volumes would be ok if they are arround 10 msec better would be < 5 as well.

Please notice if you are using NFS file systems that the network performance influences the IO counters as well.

If you are a SAP customer it makes sense to open an OSS ticket to get this analyzed in detail.(SV-BO)

This is a very common recommendation. If we have huge application traffic we need to have faster IO times then listed here.

IOf you have only peaks of such bad IO times you should try to find out what is active when you have such bad times. -> DB-Analyzer use smaller intervall.

Regards, Christiane